Podcast: Play in new window | Download
Subscribe: iTunes | Android | RSS
Justin and Jason discuss how Rackspace went down for 7 hours and took Pluggio down with it, some ideas for implementing a robust fail-over solution, the latest on AnyFu and some feedback from the first two users, why Justin is moving to Savannah, GA and what makes him think he’s actually going to get work done on the drive there, what James Dyson wants to do to improve the British economy, how Jason ghost wrote the “Help a hacker out” post on Hacker News, what happened when David Wasmer increased his “luck surface area”, what to do about the asteroid that’s headed directly at earth, why we’re underestimating the risk of human extinction and the three propositions of the ancestor simulation hypothesis, how you can double your brain power through “transcranial direct-current stimulation” and for only $99, what’s been learned from the hacking of Stratfor, why kids born later in the year are more likely to be diagnosed with ADHD, Jason’s idea for a real-word “like” system, the difference between grinds and princes, an Appignite update, and why YOU need to review us on iTunes.
The “Double Your Brain Power” link is empty.
@Steve French – Thanks, I just fixed it.
Sorry guys cannot do an iTunes review. I refuse to install that software. If there is any other way I can help out you totally know I will but I draw the line at iTunes.
@Ben Boyter – No problem, we each have our own red line. 😉
Justin,
If you want to keep a copy of the db safe for rehosting pluggio then in my opinion the safest and easiest way to configure this is to configure MySQL replication. It is trivial to set up and point it at another cheap VPS elsewhere that is waiting to pick it up. Its probably best to have it go over SSH.
You mentioned logging queries e.t.c. That is how MySQL replication works. It stores a binary log of queries or row changes depending on how you configure it and your MySQL version.
Stew
Re: Justin’s concerns about people using AnyFu, then doing business amongst themselves off the site, I agree with Jason 100%.
It’s up to you to make AnyFu so good/convenient/attractive that people prefer to do their second, third, etc. interactions there rather than elsewhere.
I don’t know how you do that, though…
@Ben Boyter, @Jason – I agree with Ben. If I get a Mac product there is something very wrong. Where else can your podcast be rated?
@Justin & Jason: Something has been bugging me about all your AnyFu discussions which I would love it if you could address for me.
The way the discussion usually goes is that Justin will either ask Jason a question or make a suggestion about how something on AnyFu should work. Nine times out of ten Jason will then proceed to explain why Justin is apparently wrong! Most amusing. Justin often quickly agrees with Jason’s reasoning, especially when the answers are “common sense”. All very good and nice to hear the discussion and the thought process.
However, it often occurs to me that a lot of Jason’s reasoning is simply not empirically backed-up. You appear to be making a lot of assumptions about AnyFu and I worry that you haven’t made any attempts to ‘test your hypotheses’. For reference I am borrowing heavily from the “Lean Startup” way of thinking which stresses the importance of making sure you can test your assumptions so that you can separate the wheat from the chaff.
So I would like to know how you are testing your hypotheses? How are you validating your assumptions? Or do you hoping that “if you build it, they will come?”
Regarding Manufacturing – you should check out Tyler Cowen’s “The Great Stagnation”, he goes into great detail about the difference in between the manufacturing vs service economy (short version – a manufacturing economy forces honesty in measurement) well worth checking out.
@Alex. You may also notice 1-3 months later Jason often comes round to may way of thinking.
@Alex Gemmell – At this stage, AnyFu is entirely an untested hypothesis and the disagreement you’re hearing is just us sorting out what that hypothesis is going to be. The real testing and validation will come once we have something that actually works.
@Justin – Sure, if believing that makes you feel any better. 😉
Don’t know if you guys have seen this nod to your podcast http://blog.liamcavanagh.com/2012/02/what-i-learned-building-a-startup-on-microsoft-cloud-services-part-1-introduction/
For the record, it was a genuine question because I’m always learning about agile/lean startup stuff – that’s one of the main reasons I listen to TZ. I wanted to know if AnyFu was following a lean style plan (sounds like a diet!) and if so could they elaborate on bits to do with hypotheses. But as Jason replied above: they’re not! So that’s the end of that!
@justin concerns:
A. You can’t charge “derivative value”. Same thing if the consulting work lead to an exit or options plan or whatever.
B. Really, if Any-fu becomes the ideal platform for head-hunting? You will disrupt a lucrative market and charge a nice commission on A LOT of transactions. What can you ask for more!
Great show guys! One of the funnier episodes 🙂
“Not rude, just an American”, “Freaked out cat and litter box on a back seat”, grinds and princes….
I was listening to it during drive to work and I can tell you – when I finally got to the office, there was no one in a better mood than me 🙂
Thanks guys!
PS. @Jason I admire your reasoning about letting people do what they want after using AnyFu… It just seems fair and you can not stop it even if you try 😉
PPS. Sorry, Itunes is a deal breaker for me to… Would have given you a heart turning, raise from the dead, give me your daughter kind of podcast review ! 😉
Hi guys,
i’ve been following you for the last months and I always like your podcast, but this time I’d like to criticize you a bit: the first is specific: Justin, please dont reinvent the wheel (manually ship a query log to another server? please use available replication solutions).
The second one is more general about this kind of down problem: all the cloud “buzz” is about having highly scalable, redundant,backed up solutions managed by big names (Amazon, Rackspace, Microsoft,Google etc..). Everything is advertised as a way to reduce your system administrations tasks (“you dont have to take care of hardware, network in IaaS or even os and platforms in PaaS”) so you dont have to hit your head trying to figure out how to have a solid infrastructure, And *you pay* for this. If your provider is not able to achieve this kind of availability, you have SLA (ask credits) or you can move to another, hopefully, better one. Of course you dont know if others are better (just look at the last year Amazon down, or the recent Azure 29th february one). I dont think it’s a good idea to choose this road, and then try to do some kind of patchwork in order to have an additional availability layer over..
If you dont trust this kind of solutions, maybe it’s better to get off “the cloud” and start planning carefully your infrastructure (servers in different datacenters, maybe self owned and managed, a dedicated Autonomous System with multiple ISP connections etcetera etcetera). But can you afford it? 🙂
@Vladimir Jankovic – I’m glad you found the show entertaining. 😉 We had a good time as well.
@Ecio – You make a very good point about cloud vs VPS hosting. While I haven’t had enough experience with cloud hosting to form an informed opinion, from the articles I’ve read on the subject I think you’re probably correct.
Also, I think @Stewart Robinson’s prescription (which is inline with your advice) sounds like the simplest and most affordable way to implement a robust failover solution for a MySQL-based application.
Just found you guys about a month ago. Working my way back now. Great stuff.
Just wanted to mention that Jason’s ‘surface area of luck’ reminds me of Taleb’s:
“This makes living in big cities invaluable because you increase the odds of serendipitous encounters — you gain exposure to the envelope of serendipity.” (From The Black Swan)
Especially in light of all the hoopla around Codecademy etc. would you consider doing an episode around learning how to code?
Best,
JohnH
Hey Guys,
Thanks for another great installment. I really enjoy the banter and more so the interviews. Any planned for next show?
I was a little surprised by the discussion this episode around Anyfu. I haven’t listened to the early episodes but what formal brainstorming and idea evaluation have you undertaken to develop the product?
I kinda agree with Justin’s questioning on customers side stepping AnyFu and arranging agreements directly. While you may have made some money from the initial meeting, you would lose any future income from later work carried out. This is surely why eLance have the exit fee Justin described? I think the negative consequences Jason was implying may not be as bad as you think, especially if you customers are using your site to get access to really great people.
Just some thoughts. Keep up the good work.
Hi guys, you are doing a great job on the show. About Jason’s idea for the real world looks like something foursquare is already doing.
here is a link : http://www.youtube.com/watch?v=t8dmjoqOOQo
keep-up the good work.
Just a little note on the PHP session stuff: I have completely stopped using PHP’s built in sessions for everything I do.
1.) Store your sessions in MySQL and then query for the cookie on every request, once you hit critical mass move the session table to Redis or Memcached.
2.) Just use on cookie for session and remember me and set the expiration in your database and for the cookie out say two weeks.
The other advantage of storing your sessions in MySQL is that you can run round-robin load balancing and the sessions are in once place.
Kind of related to the discussion you guys had about China. I highly reccommend “reverse outsourcing”. Outsource yourself to South-East Asia and telecommute/freelance for Western wages.
Benefits:
* Live like a king
* Yellow fever!
I must violently agree with Justin regarding lock-in.
Imagine I come along, and find an expert to help me sort my toe-nail clippings. I pay for an hour and we chat. To facilitate that chat, we’ve stepped outside AnyFu’s walled garden and into Skype or email or the phone system.
I get my issue sorted out and carry on my merry way. The invoice is paid and everyone is happy.
But avast! I’ve now realised I need to deal with a brand new issue. And I know my expert would be able to help me. And it’s just the tiniest of questions. It will only take him 30 seconds to reply to an email. Oh and he’s on Skype right now!
So I send him an email or phone his Skype. He replies. But that leads to another question. So, being the nice guy I am, I offer to pay him the same consult fee I paid the first time.
We spend another hour on the phone and I pay him and .. hang on .. not everyone is happy. The client and expert are happy. But you’re not. AnyFu just missed out on revenue that is rightfully theirs.
You talked about only being interested in the short consults, but you will only get the first consult. After that, any pairing you introduced will go offline.
Effectively your service is now an introduction service where the price of introduction is whatever time/rate the initial consult runs to.
If this is your goal, then charge an introduction fee rather than a consult fee. The client wants an experts time, they request an introduction, the expert agrees and AnyFu gets some cash. After the introduction fee, all monies are the responsibilities of the introduced pair.
* * *
Second scenario: The expert is contacted after the first consult, but this time he feels that it’s an imposition. He’ll respond and then wonder if AnyFu was worth the effort and not take up any further clients. Or maybe he’ll ignore the request. Or maybe he’ll respond tersely. None of these is a good outcome for AnyFu.
Imagine instead if he was required by the terms of AnyFu to insist that all small consulting contact post the initial discussion goes through AnyFu. “Hey, I don’t want to be rude, but AnyFu requires me to get you to go back through them. If I don’t, they can terminate my account”
Include in the agreement that this requirement is about short consults only. Offers of extended work (define it as .. anything longer than four hours?) aren’t within the scope of AnyFu and clients and experts are free to come to arrangements outside that.
It’s not worth suing anyone, and it’s certainly not worth $1500 to break the agreement. So just make the agreement sensible and make it so the expert is protected. After all your experts are your most precious resource.